database-scout

Pass

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a Python script scripts/db_explorer.py that handles command-line arguments and database connection strings to interact with databases.
  • [DATA_EXFILTRATION]: The skill's primary function is to read and preview database contents, which provides a direct mechanism for an agent to access and potentially exfiltrate sensitive information stored in user databases.
  • [EXTERNAL_DOWNLOADS]: The documentation references the installation of the psycopg2-binary package from the public Python Package Index (PyPI) to enable PostgreSQL support.
  • [PROMPT_INJECTION]: The scripts/db_explorer.py script employs a regex-based blacklist to filter dangerous SQL keywords. Blacklist-based security is often bypassable through dialect-specific syntax variations or sophisticated encoding techniques that may evade the defined patterns.
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface: The skill ingests untrusted data from database rows which could contain malicious instructions designed to manipulate the agent's behavior.
  • Ingestion points: Database query and preview results processed in scripts/db_explorer.py and returned to the agent.
  • Boundary markers: Results are returned in structured JSON format, but there are no explicit instructions or delimiters telling the agent to ignore embedded commands within the data.
  • Capability inventory: The script performs local filesystem reads (SQLite) and establishes network connections (PostgreSQL).
  • Sanitization: No filtering or sanitization of the database content is performed before it is presented to the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 13, 2026, 02:07 AM
Security Audit — agent-trust-hub — database-scout